svn commit: r238148 - in head/sys: conf dev/e1000 modules/igb
John Baldwin
jhb at freebsd.org
Thu Jul 5 21:32:26 UTC 2012
On Thursday, July 05, 2012 4:26:58 pm Jack F Vogel wrote:
> Author: jfv
> Date: Thu Jul 5 20:26:57 2012
> New Revision: 238148
> URL: http://svn.freebsd.org/changeset/base/238148
>
> Log:
> Sync with Intel internal source:
> shared code update and small changes in core required
> Add support for new i210/i211 devices
> Improve queue calculation based on mac type
>
> MFC after:5 days
I only spot two small regressions in this merge:
Index: if_igb.c
===================================================================
--- if_igb.c (revision 238148)
+++ if_igb.c (working copy)
@@ -767,8 +767,6 @@ igb_detach(device_t dev)
if (adapter->vlan_detach != NULL)
EVENTHANDLER_DEREGISTER(vlan_unconfig, adapter->vlan_detach);
- ether_ifdetach(adapter->ifp);
-
callout_drain(&adapter->timer);
#ifdef DEV_NETMAP
@@ -965,7 +963,7 @@ igb_mq_start(struct ifnet *ifp, struct mbuf *m)
IGB_TX_UNLOCK(txr);
} else {
err = drbr_enqueue(ifp, txr->br, m);
- taskqueue_enqueue(que->tq, &que->que_task);
+ taskqueue_enqueue(que->tq, &txr->txq_task);
}
return (err);
(The current driver calls ether_ifdetach() twice, and the second change
refixes the issue with packets being delivered out-of-order.)
However, the changes to add a (int *)cast to eee_disable are not safe.
You will need to add some sort of SYSCTL_PROC wrapper that sets the
boolean directly instead. I don't think it is safe to rely on bool
being an int.
--
John Baldwin
More information about the svn-src-all
mailing list